Conversational Entity Linking (ConEL) dataset

ConEL数据集是由奈梅亨大学创建，专注于对话系统中的实体链接任务。该数据集包含100个注释对话，总计708个用户话语，来源于QuAC、MultiWOZ、WoW和TREC CAsT 2020数据集。数据集的创建过程涉及使用众包方法对对话进行详细分析和注释，以识别概念、命名实体和个人实体。ConEL数据集的应用领域主要集中在改进对话系统中的自然语言理解，特别是在处理用户话语中的实体提及时，旨在提高系统的理解能力和交互质量。

The ConEL dataset was created by Radboud University Nijmegen, focusing on the entity linking task in dialogue systems. It contains 100 annotated dialogues, totaling 708 user utterances, which are sourced from the QuAC, MultiWOZ, WoW and TREC CAsT 2020 datasets. The dataset construction process adopts crowdsourcing methods to conduct detailed analysis and annotation on dialogues, aiming to identify concepts, named entities and personal entities. The main application scenarios of the ConEL dataset focus on improving natural language understanding in dialogue systems, especially when handling entity mentions in user utterances, with the goal of enhancing the system's comprehension ability and interaction quality.